PDF Archive

Easily share your PDF documents with your contacts, on the Web and Social Networks.

Share a file Manage my documents Convert Recover Search Help Contact



GenderDysphoria .pdf



Original filename: GenderDysphoria.pdf

This PDF 1.6 document has been generated by / PDFlib+PDI 7.0.2 (PHP5/Win32), and has been sent on pdf-archive.com on 10/09/2015 at 20:42, from IP address 170.140.x.x. The current document download page has been viewed 578 times.
File size: 86 KB (12 pages).
Privacy: public file




Download original PDF file









Document preview


Developmental Psychology
2008, Vol. 44, No. 1, 34 – 45

Copyright 2008 by the American Psychological Association
0012-1649/08/$12.00 DOI: 10.1037/0012-1649.44.1.34

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

A Follow-Up Study of Girls With Gender Identity Disorder
Kelley D. Drummond

Susan J. Bradley

Ontario Institute for Studies in Education of the University of
Toronto

Centre for Addiction and Mental Health

Michele Peterson-Badali

Kenneth J. Zucker

Ontario Institute for Studies in Education of the University of
Toronto

Centre for Addiction and Mental Health

This study provided information on the natural histories of 25 girls with gender identity disorder (GID).
Standardized assessment data in childhood (mean age, 8.88 years; range, 3–12 years) and at follow-up
(mean age, 23.24 years; range, 15–36 years) were used to evaluate gender identity and sexual orientation.
At the assessment in childhood, 60% of the girls met the Diagnostic and Statistical Manual of Mental
Disorders criteria for GID, and 40% were subthreshold for the diagnosis. At follow-up, 3 participants
(12%) were judged to have GID or gender dysphoria. Regarding sexual orientation, 8 participants (32%)
were classified as bisexual/homosexual in fantasy, and 6 (24%) were classified as bisexual/homosexual
in behavior. The remaining participants were classified as either heterosexual or asexual. The rates of
GID persistence and bisexual/homosexual sexual orientation were substantially higher than base rates in
the general female population derived from epidemiological or survey studies. There was some evidence
of a “dosage” effect, with girls who were more cross-sex typed in their childhood behavior more likely
to be gender dysphoric at follow-up and more likely to have been classified as bisexual/homosexual in
behavior (but not in fantasy).
Keywords: gender identity disorder, gender identity, sexual orientation, girls, follow-up

der identity, and subsequent gender identity and sexual orientation
in late adolescent girls and young adult women.
Several lines of evidence suggest that there are empirical reasons to posit a link between sex-typed behavior in childhood and
later gender identity and sexual orientation. Like sex-typed behavior in childhood, gender identity and sexual orientation in adulthood are also sex dimorphic: Most women have a “female” gender
identity (the subjective sense of self as a woman) and are sexually
attracted to men, whereas most men have a “male” gender identity
and are sexually attracted to women. Indeed, gender identity and
sexual orientation may be the two behavioral traits that most
strongly differentiate women from men (cf. Hyde, 2005). Using a
self-report questionnaire designed to measure gender identity dimensionally in adolescents and adults, for example, Deogracias et
al. (2007) obtained a between-sex effect size, using Cohen’s d, of
13.24.
Over the past several decades, the empirical literature has relied
on two methods, namely, retrospective and prospective designs
using targeted samples, to examine the relation between sex-typed
behavior in childhood and subsequent gender identity and sexual
orientation in adulthood. Retrospective designs have studied adults
with known variation in their gender identity and/or sexual orientation. For example, adults who meet the Diagnostic and Statistical
Manual of Mental Disorders (DSM) criteria for gender identity disorder (GID; also known as transsexualism) recall engaging in more
cross-gender-typed behavior in childhood than do adults without GID
(e.g., Blanchard & Freund, 1983; Doorn, Poortinga, & Verschoor,
1994; Ehrhardt, Grisanti, & McCauley, 1979; Freund, Langevin,
Satterberg, & Steiner, 1977; see also Bartlett & Vasey, 2006).

Research on normative (or typical) gender development has
documented various behavioral domains in which children show,
on average, significant sex differences: gender identity selflabeling, sex-of-playmate preference, toy and activity interests,
roles in fantasy play, parental rehearsal play, and so on (for a
review, see Ruble, Martin, & Berenbaum, 2006; Zucker, 2005c).
The determinants of this between-sex variation in sex-typed behavior have long been deemed by developmentalists to have important implications for other aspects of psychosocial development, such as interpersonal relational styles (e.g., Maccoby, 1998),
cognitive skills (e.g., Liss, 1983), and vocational interests (e.g.,
Lippa, 1998), for which there are also significant sex differences.
As noted by Lippa (2002), determining within-sex individual
differences in gender-related behavior is another strategy used to
study variations with regard to other aspects of development (see,
e.g., Barrett & White, 2002; Khuri & Ruble, 2006). In the present
study, we used this approach to examine the relation, if any,
between sex-typed behavior patterns in childhood, including gen-

Kelley D. Drummond and Michele Peterson-Badali, Department of
Human Development and Applied Psychology, Ontario Institute for Studies in Education of the University of Toronto, Toronto, Ontario, Canada;
Susan J. Bradley and Kenneth J. Zucker, Gender Identity Service, Child,
Youth, and Family Program, Centre for Addiction and Mental Health,
Toronto, Ontario, Canada.
Correspondence concerning this article should be addressed to Kenneth
J. Zucker, Gender Identity Service, Child, Youth, and Family Program,
Centre for Addiction and Mental Health, 250 College Street, Toronto,
Ontario M5T 1R8, Canada. E-mail: ken_zucker@camh.net
34

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

SPECIAL SECTION: GENDER IDENTITY DISORDER IN GIRLS

The largest body of retrospective research pertains to the withinsex association between sex-typed behavior in childhood and
sexual orientation in adulthood. Bailey and Zucker (1995) performed a meta-analysis of 41 retrospective studies that made a
quantitative comparison between heterosexual and homosexual
same-sex adults using some measure of childhood sex-typed behavior. These studies yielded 48 independent effect sizes: 32
compared heterosexual and homosexual men, and 16 compared
heterosexual and homosexual women. Using Cohen’s d, Bailey
and Zucker found that there were substantial differences in patterns of recalled childhood sex-typed behavior between heterosexual and homosexual adults. On average, both homosexual men and
women recalled more cross-sex-typed behavior in childhood than
did their heterosexual counterparts (respective ds were 1.31 and
0.96). Subsequent studies have, with no exception, continued to
replicate these findings (summarized in Zucker et al., 2006).
There are, of course, both methodological and interpretive problems with retrospective designs (for an overview, see Hardt &
Rutter, 2004). In a targeted sample of adults with GID (invariably
recruited from specialized gender identity clinics), it is possible
that the association with cross-sex-typed behavior is magnified
because not all individuals with pervasive cross-gender behavior in
childhood end up seeking out medically assisted gender change in
adulthood (e.g., because their earlier gender dysphoria had desisted). In the studies comparing the recollections of heterosexual
and homosexual adults, in which there is less of a sampling bias
problem, the most common criticism has pertained to memory
distortion or selective recall. For example, it has been argued that
the greater recollection of cross-gender behavior during childhood
by homosexual than by heterosexual adults is linked to the widespread “master narrative” in Western culture that presupposes that
“gender inversion” is linked to homosexual sexual orientation (see,
e.g., Cohler & Galatzer-Levy, 2000; Gottschalk, 2003; Hegarty,
1999; Kite & Deaux, 1987). As a result, it has been claimed that
the sex-typed behavior–sexual orientation association is nothing
more than participants recalling behaviors that adhere to cultural
stereotypes and expectations. Although there is evidence that
speaks against this retrospective distortion hypothesis (summarized in Bailey & Zucker, 1995; Zucker, 2005a, in press; Zucker et
al., 2006), there is general agreement that the retrospective data
should be confirmed (or disconfirmed) with prospective designs.
One prospective approach has been to target a sample of children presumed to have moderate-to-pervasive cross-gender behavior. In one line of research, sampling consisted of ad-recruited girls
with parent-nominated “tomboyish” behavior, along with measures of sex-typed behavior administered to the girls themselves
(e.g., Bailey, Bechtold, & Berenbaum, 2002; Berenbaum & Bailey,
2003; Green, Williams, & Goodman, 1982), who were compared
to girls unselected for their gender behavior. Neither research team
has, as of yet, reported on longer term linkages.
A second strategy has been to study children referred to specialized gender identity clinics because there is concern about their
cross-gender behavior and gender identity status (e.g., on the part
of parents, mental health professionals, teachers, etc.). Over the
years, several research teams have studied such children, and
overviews may be found in the work of Green (1987), Zucker and
Bradley (1995), and Cohen-Kettenis and Pfa¨fflin (2003).
In one study, Green (1987) assessed the gender identity and
sexual orientation of 44 behaviorally feminine boys and 30 control

35

boys who were at a follow-up mean age of 18.9 years (range,
14 –24 years) and who had initially been evaluated at a mean age
of 7.1 years (range, 4 –12 years). Of the 44 behaviorally feminine
boys, only 1 youth, at the age of 18 years, was gender dysphoric
to the extent of considering sex-reassignment surgery. None of the
other boys were reported to have gender identity problems at
follow-up. Sexual orientation in fantasy and behavior was assessed
by means of a semistructured, face-to-face interview. Kinsey ratings were made on a 7-point continuum, ranging from exclusive
heterosexuality (a Kinsey “0”) to exclusive homosexuality (a Kinsey “6”; Kinsey, Pomeroy, & Martin, 1948). Depending on the
measure (fantasy or behavior), 75%– 80% of the previously behaviorally feminine boys were either bisexual or homosexual
(Kinsey ratings between 2 and 6) at follow-up versus 0%– 4% of
the control boys.
Data from seven other follow-up reports on a total of 82 behaviorally feminine boys have been summarized in detail elsewhere
(Zucker, 2005b; Zucker & Bradley, 1995, pp. 285–286, 290 –297).
Similar to Green’s (1987) case-control study, these studies also
identified an elevated rate of either a bisexual or homosexual
sexual orientation (52.4%). In contrast to Green’s (1987) study,
however, the other studies found the rate of GID persistence was
higher, with rates ranging from 12% to 20%.
From these prospective studies of behaviorally feminine boys,
two conclusions might be drawn: (a) The rate of persistent gender
dysphoria was modest but arguably higher than one estimated base
rate for gender dysphoria in the general population of biological
males: 1 in 11,000 men (Bakker, van Kesteren, Gooren, & Bezemer, 1993), and (b) the rate of a later bisexual or homosexual
sexual orientation was notably higher than the known base rates
for a bisexual or homosexual sexual orientation in the general
population of biological males (see, e.g., Laumann, Gagnon, Michael, & Michaels, 1994). Thus, for sexual orientation, there
appears to be a reasonable convergence between prospective and
retrospective studies but, for gender identity, there is more divergence: Many boys with pervasive cross-gender behavior and cooccurring gender dysphoria do not show persistent gender dysphoria by late adolescence or young adulthood, which is at some
variance from the recollections of most gender-dysphoric adolescent boys and adult men.
Over the years, it has been noted that little is known about the
longer term psychosexual outcome of girls referred to specialized
gender identity clinics (Peplau & Huppin, in press; Peplau, Spalding, Conley, & Veniegas, 1999). In part, this has been a function
of the fact that boys are much more likely than girls to be referred
to gender identity clinics: 5.75:1 in one clinic and 3.07:1 in another
(see Cohen-Kettenis, Owen, Kaijser, Bradley, & Zucker, 2003;
Cohen-Kettenis et al., 2006). The present study attempted to fill
this gap by providing, to our knowledge, the first systematic
follow-up report of clinic-referred girls with GID with regard to
gender identity and sexual orientation.

Method
Participants
Between 1975 and 2004, 71 girls (age range, 3–12 years) were
referred for assessment to the Gender Identity Service, Child,
Youth, and Family Program at the Centre for Addiction and

DRUMMOND, BRADLEY, PETERSON-BADALI, AND ZUCKER

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

36

Mental Health in Toronto, Ontario, Canada. To participate in the
follow-up study, patients had to be at least 17 years of age. Using
this age cutoff, we identified 37 eligible girls, of whom 30 were
contacted for participation. Of the remaining 7 girls, 3 could not be
traced through previous addresses, registrars, and personal contacts (e.g., the patient and/or their family had moved and a current
telephone number, mailing address, or e-mail address could not be
identified), and 4 were not available to participate within the time
requirements of the study.
Initial telephone contact was first made with the parents or legal
guardians because participants were minors at the time of assessment and some may have had no recollection of their clinic
attendance.1 Of the 30 clients contacted, 25 (83.3%) agreed to
participate; 24 came into the clinic for testing, and 1 participant
completed a telephone interview because she was too anxious to
travel to the clinic. Of the remaining 5 girls, 4 of the girls’ parents
or guardians (e.g., the Children’s Aid Society) were unwilling to
provide contact information for their children. One individual
declined to participate.
The demographic characteristics of the participants in childhood
and at follow-up are shown in Table 1. The GID diagnosis in
childhood was based on the DSM (3rd ed. [DSM–III]; 3rd ed., rev.
[DSM–III–R]; or 4th ed. [DSM–IV]; American Psychiatric Association [APA], 1980, 1987, and 1994, respectively) criteria applicable at the time of assessment. Fifteen girls (60%) met complete

Table 1
Demographic Characteristics (N ⫽ 25)
Characteristic
From childhood
Age (in years)
Year of assessment
IQa, b
Social classc
Marital statusd
Two-parent family
Other
Caucasian
At follow-up
Age (in years)e
Year of birth
Interval (in years)f
IQb, g
a

M

SD

Range

8.88
1989.36
105.17
35.72

3.10
7.02
21.73
14.40

3.17–12.95
1977–2002
57–144
8–66

%

Procedure

60.0
40.0
80.0
23.24
1980.52
14.34
10.20

5.82
6.06
7.03
2.71

DSM criteria for GID in childhood. The remaining 40% were
subthreshold for a DSM diagnosis of GID, but all had some
indicators of GID, and some would have met the complete DSM
criteria at some point in their lives prior to their assessment in
childhood.
Four of the girls in the follow-up sample were born with a
disorder of sex development (DSD; 2 had cloacal exstrophy, 1 had
congenital micropenis syndrome of unknown etiology, and 1 had
mixed gonadal dysgenesis; Hughes, Houk, Ahmed, Lee, & Lawson Wilkins Pediatric Endocrine Society/European Society for
Paediatric Endocrinology Consensus Group, 2006). Three of the
nonparticipants also had a DSD (partial androgen insensitivity
syndrome, congenital adrenal hyperplasia, or true hermaphroditism). There are arguments for and against the inclusion of the 4
girls with a DSD in this sample (see, e.g., Meyer-Bahlburg, 1994).
A female gender assignment was made for all 4 girls almost
immediately after birth. Also in early infancy, the 4 girls were
gonadectomized and had surgical feminization of their external
genitalia. Like the somatically intact girls, the 4 girls were referred
for concern about their gender development in relation to their
assigned gender. On the one hand, as noted by Meyer-Bahlburg
(2005), “there is every reason to assume that the processes and
psychosocial factors involved in normative gender development
also contribute to development of all aspects of gender. . .in persons with intersexuality” (p. 434). On the other hand, as also noted
by Meyer-Bahlburg (2005), “additional factors. . .may come into
play in [such persons]. . .particularly the awareness of an atypical
biological condition and medical history” (pp. 434 – 435). As noted
in Table 3, only 1 of these girls met the complete Point A and Point
B DSM criteria for GID, and the other 3 were subthreshold.

15.44–36.58
1968–1989
2.99–27.12
5.00–15.75

Full-scale IQ was obtained with age-appropriate Wechsler intelligence
scales (the Wechsler Preschool and Primary Scale of Intelligence—Third
Edition [Wechsler, 2002], the Wechsler Intelligence Scale for Children—
Revised [Wechsler, 1974], and the Wechsler Intelligence Scale for Children—Third Edition [Wechsler, 1991]). One participant was administered
the Stanford-Binet Intelligence Scale (Thorndike, Hagen, & Sattler,
1986). b IQ scores at assessment and follow-up were not available for 1
participant. c For social class, Hollingshead’s (1975) Four Factor Index
of Social Status was used. The absolute range was 8 – 66. d For marital
status, the category “Other” included the following family constellations:
single parent, separated, divorced, living with relatives, or in the care of the
Children’s Aid Society. e One participant (who was 15.44 years of age)
was below the lower bound age cutoff of 17 years but was included in the
study because her guardian had contacted the clinic for issues unrelated to
gender identity status. f Interval denotes the time between childhood
assessment and follow-up assessment. g Composite IQ ⫽ (Vocabulary ⫹
Comprehension ⫹ Block Design ⫹ Object Assembly subscale scores)/4.
The absolute range was 1–19.

All participants were evaluated on a single day. Below, we
provide information on the measures used in this report (for other
measures, including parent and self-ratings of behavior problems,
psychiatric diagnoses, and experiences of stigma, see Drummond,
2006). All of the participants provided written informed consent
prior to their involvement in the follow-up assessment and were
provided a stipend for their participation and reimbursement for
travel expenses. The study was approved by the Institutional
Review Boards at the Centre for Addiction and Mental Health and
the University of Toronto.

1

It is beyond the scope of this report to describe the types of therapies
(as well as their frequency and duration) that the girls and/or their parents
may have received between the assessment in childhood and the follow-up
(e.g., by a therapist within the Gender Identity Service at the Centre for
Addiction and Mental Health or in the community). From the participants’
clinic files, 13 of the 25 girls had at least some contact with our clinic
during the interval between assessment and follow-up (e.g., as therapy
clients or for a reassessment). Of the 25 girls and/or their parents, 18 had
been in some type of therapy or counseling during the interval between
assessment and follow-up; of these, 5 were patients of staff within the
Gender Identity Service, and the remainder were seen by a professional in
the community.

SPECIAL SECTION: GENDER IDENTITY DISORDER IN GIRLS

Measures

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Childhood Assessment
Cognitive functioning. IQ was assessed with the Wechsler
Adult Intelligence Scale—Third Edition (Wechsler, 1997) or the
Wechsler Intelligence Scale for Children—Third Edition (Wechsler, 1991) and, for one participant, with the Stanford-Binet Intelligence Scale (Thorndike, Hagen, & Sattler, 1986).
Sex-typed behavior. Five child informant and three parent
informant measures were used to assess the participants’ sex-typed
behavior in childhood: (a) the Draw-a-Person test (Zucker, Finegan, Doering, & Bradley, 1983); (b) a free-play task (Zucker,
Doering, Bradley, & Finegan, 1982); (c) the Playmate and Playstyle Preferences Structured Interview (Fridell, Owen-Anderson,
Johnson, Bradley, & Zucker, 2006); (d) sex-typed responses on the
Rorschach test (Zucker, Lozinski, Bradley, & Doering, 1992); (e)
the Gender Identity Interview (Zucker et al., 1993); (f) the Gender
Identity Questionnaire for Children (Johnson et al., 2004); (g) a
measure of activity level/extraversion (Zucker & Bradley, 1995);
and (h) the Games Inventory (Bates & Bentler, 1973). These child
and parent informant measures all had established discriminant
validity, that is, they significantly differentiated the clinic girls
referred for gender identity concerns from control girls (for a
review, see Zucker, 2005c; Zucker & Bradley, 1995). A Childhood
Sex-Typed Behavior Composite was computed for each participant by averaging the z-scores for these measures (which yielded
a total of 11 indices), as well as the GID DSM diagnosis (1 ⫽
threshold, 2 ⫽ subthreshold) in childhood. Data from the total
sample of participants and nonparticipants (N ⫽ 37) were used.
Because of missing data, the mean number of indices/participant
was 9.16 (SD ⫽ 2.30).

Follow-Up Assessment
Cognitive functioning. Four subtests (Vocabulary, Comprehension, Block Design, and Object Assembly) of the Wechsler
Adult Intelligence Scale—Third Edition or the Wechsler Intelligence Scale for Children—Third Edition were administered. The
standard scores from the subtests were averaged to form an IQ
score for cognitive functioning.
Recalled childhood gender identity and gender role behavior.
Participants completed the Recalled Childhood Gender Identity/
Gender Role Questionnaire (RCGI; Zucker et al., 2006). This
questionnaire consists of 23 items pertaining to various aspects of
sex-typed behavior, as well as to the relative closeness to the
mother and father during childhood. Individual items were rated on
a 5-point response scale. Each participant was instructed to make
ratings for her behavior as a child (“between the years 0 to 12”).
Factor analysis identified two factors, accounting for 37.4% and
7.8% of the variance, respectively (all factor loadings ⱖ.40).
Factor 1 consisted of 18 items that pertained to childhood gender
role and gender identity, and Factor 2 consisted of three items that
pertained to parent– child relations (relative closeness to one’s
mother versus one’s father). Information on normative sex differences and discriminant validity was reported in Zucker et al.
(2006). For the present study, the mean Factor 1 score was computed for each participant.
Concurrent gender identity. During an audiotaped interview,
each participant was asked to describe her current feelings about

37

being female and then to describe positive and negative aspects
about her gender status. The examiner also asked semistructured
gender identity questions from the adolescent and adult GID
criteria outlined in the DSM–IV–TR (APA, 2000). The interviewer
asked four questions related to the Point A criteria (e.g., the stated
desire to be a man, the desire to live or to be treated as a man) and
six questions from the Point B criteria (e.g., a preoccupation with
getting rid of breasts or genitalia). Participants were asked to
respond according to the last 12 months with No, Yes, or Sometimes. Participants who answered Yes or Sometimes for one or
more of the questions from both Point A and B criteria were
classified as displaying persistent gender dysphoria.
The female version of the Gender Identity/Gender Dysphoria
Questionnaire for Adolescents and Adults (GIDQ-AA; Deogracias
et al., 2007) was also completed. This 27-item questionnaire measures gender identity and gender dysphoria in adolescents or
adults. Item content was based on prior measures, expert panels,
and clinical experience. Each item was rated on a 5-point response
scale ranging from Never to Always based on a time frame of the
past 12 months. Item examples include the following: “In the past
12 months, have you felt unhappy about being a woman?” and “In
the past 12 months, have you wished to have an operation to
change your body into a man’s (e.g., to have your breasts removed
or to have a penis made)?” Factor analysis identified a strong
one-factor solution that accounted for 61.3% of the variance. All
27 items had factor loadings ⱖ.30 (median, .86; range, .34 –.96).
Psychometric evidence for discriminant validity and clinical utility
can be found in Deogracias et al. (2007). Participants’ GIDQ-AA
total scores were calculated by summing scores on the completed
items and dividing by the number of marked responses.
Sexual orientation in fantasy. Each participant’s sexual orientation in fantasy was assessed with specific questions during an
audiotaped face-to-face interview and the self-report Erotic Response and Orientation Scale (EROS; Storms, 1980). Questions
posed in the interview addressed four types of sexual fantasy: (a)
crushes on other people, (b) sexual arousal to visual stimuli (e.g.,
to strangers, acquaintances, partners, and individuals presented in
the media [video, movies, magazines, the internet]), (c) sexual
content of night dreams, and (d) sexual content of masturbation
fantasies. Using the Kinsey scale criteria, the interviewer assigned
ratings that ranged from 0 (exclusively heterosexual) to 6 (exclusively homosexual) for each parameter. A dummy score of 7
denoted that the participant did not experience or report any
fantasies. A global fantasy score was derived on the basis of
ratings from the four questions. In the present study, only ratings
for the last 12 months are reported.
During the interview, participants were not asked directly about
the gender of the person or persons who elicited sexual arousal,
thus allowing time for the participant to provide this information
spontaneously. Directed questions were asked only if the participant did not volunteer specific information about same-sex or
opposite-sex partners. This approach was used so that, by the end
of the interview, the participant provided information about sexual
arousal to both same-sex and opposite-sex individuals.
The EROS is a 16-item self-report measure assessing sexual
orientation in fantasy over the past 12 months. Half of the questions pertained to heterosexual fantasy (e.g., “How often have you
had any sexual feelings (even the slightest) while looking at a
man?”) and the other half pertained to homosexual fantasy (e.g.,

DRUMMOND, BRADLEY, PETERSON-BADALI, AND ZUCKER

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

38

“How often have you had any sexual feelings (even the slightest)
while looking at a woman?”). Each item was rated on a 5-point
scale for frequency of occurrence, ranging from “none” to “almost
every day.” Mean homoerotic and heteroerotic fantasy scores were
derived for each participant. Previous use of the EROS has shown
good evidence of discriminant validity (Storms, 1980; Zucker et
al., 1996).
Sexual orientation in behavior. Each participant’s sexual orientation in behavior was assessed with specific questions during
the face-to-face interview and with a modified version of the
Sexual History Questionnaire (SHQ; Langevin, 1985). In the interview, questions asked about five types of sexual behavior: (a)
dating; (b) holding hands in a romantic manner; (c) kissing; (d)
genital fondling or being touched on the breasts (or, in cases of
same-sex sexual behavior, touching another woman’s breasts); and
(e) penile–vaginal intercourse, anal intercourse, or the use of
dildos. Kinsey ratings for behavior for the past 12 months were
made in the same manner as fantasy ratings.
The modified SHQ consisted of 20 questions. Ten questions
pertained to heterosexual experiences (e.g., “How many men have
you kissed on the lips in a romantic way?”), and 10 questions
pertained to homosexual experiences (e.g., “How many women
have you kissed on the lips in a romantic way?”). Each item was
rated, for the 12 month period prior to the follow-up assessment,
on a 5-point scale for frequency of occurrence, ranging from none
to 11 or more. Mean total scores for heterosexual and homosexual
experiences were derived.
Sexual identity self-labeling. Participants were asked to provide a label for their current sexual identity and were offered the
following options: (a) “straight” or “heterosexual”; (b) “lesbian,”
“homosexual,” or “queer”; (c) “bisexual”; (d) “asexual”; or (e)
“other.”
Social desirability. Social desirability can threaten the validity
of self-report scales when respondents seek social approval or try
to represent themselves in a favorable manner (King & Brunner,
2000). Participants ⱖ18 years of age completed the Marlowe–
Crowne Social Desirability Scale (M–C SDS; Crowne & Marlowe,
1960), which consists of 33 true–false items. The scale consists of
18 culturally acceptable but unlikely statements keyed in the true
direction and 15 socially undesirable but probable statements
keyed in the false direction for a maximum possible score of 33.
Participants under 18 years of age completed a shorter version of
the M–C SDS (Strahan & Gerbasi, 1972). This scale consists of 12
culturally acceptable but improbable statements keyed in the true
direction and 8 socially undesirable but probable statements keyed
in the false direction for a maximum possible score of 20. Several
studies have found that the M–C SDS is a reliable and valid
measure (Crowne & Marlowe, 1960; Holden & Fekken, 1989;
Silverthorn & Gekoski, 1995).

Table 2
Mean Factor 1 Score on the Recalled Childhood Gender
Identity/Gender Role Questionnaire (Zucker et al., 2006)
Group

M

SD

d

Total sample
(Female university students)
(Mothers of boys with GID)
(Mothers of control boys)
(Mothers of nonreferred boys)
(Sisters/female cousins of
women with CAH)
Childhood diagnosis
GID: Threshold
GID: Subthreshold

2.57
(3.43)
(3.80)
(3.72)
(3.77)

.67
(.54)
(.54)
(.34)
(.39)

20
(100)
(230)
(13)
(24)

(3.70)

(.43)

(15)

2.48
2.70

.66
.69

.32

n

11
9

Note. Absolute range is 1.00 –5.00. A lower score indicates more recalled
atypical gender identity and gender role behavior. Groups and values in
parentheses are from Zucker et al. (2006). GID ⫽ gender identity disorder;
CAH ⫽ congenital adrenal hyperplasia.

sures, it appears that the participants were representative of the
total pool of available patients and thus did not constitute a
markedly biased sample at follow-up.

Sex-Typed Behavior in Childhood
Table 2 shows the mean RCGI Factor 1 score, which pertained
to the participants’ recollections of their sex-typed behavior from
childhood. This mean score can be compared with the scores of
several samples of women, unselected for their gender identity or
sexual orientation, reported on in Zucker et al. (2006) and also
shown in Table 2. By comparing the mean factor score with the
scores from the other samples (mean range, 3.43–3.80), we see it
is apparent that the women in this study recalled relatively more
cross-gender behavior in childhood (M ⫽ 2.57, SD ⫽ .67).
Table 2 also shows the mean RCGI Factor 1 score of the
participants as a function of DSM diagnostic status in childhood.
Although the threshold participants recalled, on average, more
cross-gender behavior in childhood than the subthreshold participants, the difference was not significant, t(18) ⬍ 1; the effect size
(Cohen’s d) of .32 would be considered small. We also examined
the z-composite for childhood sex-typed behavior as a function of
diagnostic status (for this analysis, the DSM metric was removed
from the composite and served as the independent variable). With
age at assessment in childhood covaried, the threshold participants
had, on average, significantly more cross-sex-typed behavior in
childhood (M ⫽ .15, SD ⫽ .54) than did the subthreshold participants (M ⫽ ⫺.31, SD ⫽ .36), F(1, 21) ⫽ 23.36, p ⬍ .001, partial
␩2 ⫽ .53.

Results
Psychosexual Differentiation at Follow-Up
Participants Versus Nonparticipants
A preliminary analysis compared the assessment information
from childhood of the 25 girls who participated in the study with
that of the 12 girls who did not participate. There were no significant differences between the participants and nonparticipants on
any of these variables (data not shown).2 At least by these mea-

A summary of the psychosexual differentiation data, including
gender identity at follow-up, sexual orientation, and sexual identity
self-labeling for each participant, is shown in Table 3.
2

These data are available in the study by Drummond (2006).

SPECIAL SECTION: GENDER IDENTITY DISORDER IN GIRLS

39

Table 3
Summary of Gender Identity and Sexual Orientation Results at Follow-Up

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Participant
ID
1
2
3
4
5
6a
7
8
9
10
11
12
13
14
15
16
17
18
19a
20
21
22a
23
24a
25

Age at
assessment
(years)

Age at
follow-up
(years)

Global Kinsey ratings
Fantasy

9.74
8.88
5.85
3.17
4.92
5.75
12.67
12.95
8.41
8.29
4.10
4.72
6.70
6.81
12.62
12.16
7.32
8.51
12.88
9.20
11.26
12.18
12.45
11.89
8.79

36.58
36.61
32.41
28.78
26.61
26.58
17.09
28.72
23.34
24.12
20.04
19.73
21.53
18.73
23.57
21.10
17.51
17.34
21.58
17.81
19.27
17.35
15.44
27.74
23.12

6
6
0
0
4
0

6
6
4
0
0
0
0
6
6
0
0
0
0
0

0
0
0

Behavior

Sexual
identity
label

Gender identity

DSM

6
6
0

0
0


6
6
0

0
0
6
6
0


0
0


0
0

HS
HS
HT
HT
BS
HT
AS
HS
HT
BS
HT
HT
HT
HT
HS
HT
HT
HT
HT
HT
HT
HT
HT
HT
HT

WNL
WNL
WNL
WNL
WNL
WNL
Dysphoric
WNL
Dysphoric
WNL
WNL
WNL
WNL
WNL
WNL
Dysphoric
WNL
WNL
WNL
WNL
WNL
WNL
WNL
WNL
WNL


















Note. For Kinsey ratings (last 12 months), 0 ⫽ exclusively heterosexual and 6 ⫽ exclusively homosexual. In the DSM column, a plus sign indicates the
participant met complete DSM-III, DSM-III-R, or DSM-IV symptom criteria for gender identity disorder at initial assessment. Dashes indicate the participant
did not report fantasy or behavior. ID ⫽ identification label; HS ⫽ homosexual (lesbian); HT ⫽ heterosexual or straight; BS ⫽ bisexual; AS ⫽ asexual;
WNL ⫽ within normal limits (i.e., the participant did not report any distress about being a female).
a
Participant with a disorder of sex development.

Gender Identity at Follow-up
On the basis of their answers to the semistructured clinical
interview questions, participants were classified as either gender
dysphoric or not gender dysphoric. In answering these questions,
22 participants (88%) reported no distress with their female gender
identity at follow-up. None of the participants desired contrasex
hormones or sex reassignment surgery to masculinize their bodies,
nor did they express a desire to get rid of their female sex
characteristics.
The remaining 3 participants (12%) were classified as gender
dysphoric at follow-up (none of these 3 girls had a co-occurring
DSD). Among these 3 participants, 1 had been living as a boy
since early adolescence (i.e., was known to others as a boy) and
was in the process of legally changing his name on official documents. The other 2 participants were living as girls, although both
were often perceived of as boys by naı¨ve others (e.g., new acquaintances, strangers, etc.), which they preferred. All 3 gender
dysphoric participants wished they had been born a boy and
wondered whether they would have been happier as a boy. Two of
these individuals indicated a desire to have surgery to masculinize
their bodies. The other participant classified as gender dysphoric
reported indifference with regard to altering her physical appearance but felt that “it was better to be neutral.” On the basis of this
information, 2 of the participants met DSM–IV–TR criteria for
GID. Although the other participant did not meet full criteria for

GID, information from the clinical interview and semistructured GID interview indicated that she was gender dysphoric at
follow-up.
In the Deogracias et al. (2007) study, a cutoff score of ⱕ3.00
was used to indicate “caseness” for gender dysphoria on the
GIDQ-AA. The 2 participants classified as gender dysphoric (and
who completed the GIDQ-AA) had scores lower than 3.00 (means
of 2.19 and 2.26, respectively), whereas the 18 participants classified as not gender dysphoric (and who completed the GIDQ-AA)
all had scores 3.00 (M ⫽ 4.78, SD ⫽ .20; range, 4.30 –5.00). There
was a significant difference between these two subgroups, t(18) ⫽
17.81, p ⬍ .001, d ⫽ 13.27, which supports the classification of
the participants on the basis of the clinical interview.
Bakker et al. (1993) estimated that 1 in 30,400 genetically
female adults in the general population have GID. Using this
baseline prevalence value, the odds of persistent gender dysphoria
(12%) in the present sample was 4,084 times the odds of gender
dysphoria in the general population of biological females.

Sexual Orientation
On the basis of the Kinsey interview ratings, participants were
classified into the following three sexual orientation groups for
fantasy and behavior: (a) heterosexual (Kinsey ratings of 0 –1), (b)
bisexual/homosexual (Kinsey ratings of 4 – 6), and (c) no sexual
fantasy or behavior. For the fantasy ratings (see Table 3), 15

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

40

DRUMMOND, BRADLEY, PETERSON-BADALI, AND ZUCKER

participants (60%) were classified as exclusively heterosexual, 8
(32%) were classified as bisexual/homosexual, and the remaining
2 (8%) were classified as having no sexual fantasies. Of the 3
participants classified as gender dysphoric, 2 were exclusively
homosexual in fantasy (i.e., sexually attracted to members of their
own birth sex). The other gender dysphoric participant reported no
sexual fantasies and described herself as being “dead sexually.”
(Of the 4 participants with a DSD, 3 were classified as exclusively
heterosexual in fantasy, and 1 reported no sexual fantasies; 2 were
classified as exclusively heterosexual in behavior, and 2 reported
no sexual behavior.)
For the EROS, we compared the participants classified as exclusively heterosexual with those classified as bisexual/
homosexual on the basis of their Kinsey ratings. With age at
follow-up covaried, a 2 (sexual orientation: heterosexual vs. bisexual/homosexual) ⫻ 2 (EROS: attraction to men vs. attraction to
women) analysis of covariance (ANCOVA) revealed a significant
Sexual Orientation ⫻ EROS interaction, F(1, 20) ⫽ 25.67, p ⬍
.001, partial ␩2 ⫽ .56.
Independent t tests showed that participants classified as heterosexual in fantasy had, on average, a higher heteroerotic EROS
score (M ⫽ 2.03, SD ⫽ .87) than participants classified as bisexual/homosexual in fantasy (M ⫽ 1.84, SD ⫽ 1.34), but the difference was not significant, t(20) ⬍ 1, d ⫽ .19; however, participants
classified as bisexual/homosexual reported, on average, a significantly higher EROS homoerotic score (M ⫽ 3.32, SD ⫽ 1.25) than
participants classified as heterosexual (M ⫽ 1.02, SD ⫽ .07),
t(20) ⫽ ⫺7.28, p ⬍ .001, d ⫽ ⫺3.33. A paired-samples t test was
conducted to evaluate whether participants classified as heterosexual reported higher heteroerotic fantasies than homoerotic fantasies. The results indicated that the mean heteroerotic score was
significantly greater than the mean homoerotic score, t(14) ⫽ 4.75,
p ⬍ .001, with a large effect size of 1.23. Conversely, participants
classified as bisexual/homosexual reported significantly higher
homoerotic fantasies then heteroerotic fantasies, t(6) ⫽ ⫺2.61,
p ⬍ .04, with a large effect size of ⫺.99.
Regarding Kinsey ratings of sexual orientation in behavior (see
Table 3), 11 participants (44%) were classified as exclusively
heterosexual, 6 (24%) were classified as bisexual/homosexual, and
the remaining 8 (32%) were classified as having no sexual experiences. Of the 3 participants classified as gender dysphoric, 2
were exclusively homosexual in behavior (i.e., had sexual experiences with members of their own birth sex). The other gender
dysphoric participant reported no sexual behaviors.
For the SHQ ratings, we compared the participants classified as
exclusively heterosexual with those classified as bisexual/
homosexual on the basis of their Kinsey ratings. A 2 (sexual
orientation: heterosexual vs. bisexual/homosexual) ⫻ 2 (SHQ:
with men vs. with women) analysis of variance (ANOVA) revealed a significant Sexual Orientation ⫻ SHQ interaction, F(1,
13) ⫽ 70.41, p ⬍ .001, partial ␩2 ⫽ .84. Independent t tests for the
SHQ scores showed that participants classified as heterosexual in
behavior reported, on average, significantly more heterosexual
sexual experiences (M ⫽ 2.15, SD ⫽ .54) than participants classified as bisexual/homosexual (M ⫽ 1.00, SD ⫽ .00), t(13) ⫽ 4.12,
p ⫽ .001, d ⫽ 2.42. In fact, participants classified as bisexual/
homosexual reported no sexual experiences with men over the past
12 months. Participants classified as bisexual/homosexual reported, on average, significantly more homosexual sexual experi-

ences (M ⫽ 2.48, SD ⫽ .40) than did participants classified as
heterosexual (M ⫽ 1.04, SD ⫽ .12), t(13) ⫽ ⫺11.17, p ⬍ .001,
d ⫽ ⫺6.56.
For participants classified as having a “typical” (i.e., nongender-dysphoric) gender identity at follow-up, there were no
substantive disjunctions between Kinsey ratings and sexual identity self-labeling (see Table 3). One exception was a participant
who self-labeled as heterosexual, although she did not report any
sexual fantasies or behaviors in the 12 months prior to the interview. For the 3 participants classified as gender dysphoric at
follow-up, 2 self-labeled as heterosexual; however, it should be
noted that their sexual orientation in relation to their birth sex was
homosexual. As noted earlier, the remaining gender-dysphoric
participant felt that she was “dead sexually” and labeled herself as
asexual.
One participant (ID 5 in Table 3) was classified as bisexual/
homosexual in fantasy but heterosexual in behavior. Her selflabeled sexual identity was bisexual. For the 17 participants who
could be assigned a Kinsey rating between 0 and 6 for both
behavior and fantasy (i.e., excluding the 8 individuals who did not
report any sexual behavior [n ⫽ 6] or any sexual fantasy and
behavior [n ⫽ 2]; see Table 3), the correlation between Kinsey
fantasy and behavior ratings was .93 (df ⫽ 15), p ⬍ .001.

Odds Ratios for Bisexual/Homosexual Sexual Orientation
in Fantasy and Behavior
Odds ratios were calculated for bisexual/homosexual sexual
orientation in fantasy and behavior using prevalence estimates
from several major survey studies of sexual orientation in adolescent girls and young women (Dickson, Paul, & Herbison, 2003;
Fergusson, Horwood, Ridder, & Beautrais, 2005; McCabe,
Hughes, Bostwick, & Boyd, 2005; Narring, Stronski, & Michaud,
2003; Remafedi, Resnick, Blum, & Harris, 1992; Russell & Seif,
2002). From these studies, base rates for bisexual/homosexual
sexual orientation in fantasy and behavior were estimated to range
from 2.0% to 5.0% in the female general population. The odds of
reporting bisexual/homosexual sexual orientation in fantasy in the
present sample was 8.9 –23.1 times higher, and the odds of reporting bisexual/homosexual sexual orientation in behavior in the
present sample was 6.0 –15.5 times higher than it is in women in
the general population.

Relation Between Age and Sexual Orientation
Table 4 shows the means and standard deviations of ages at
assessment and at follow-up as a function of Kinsey groups in
fantasy and behavior, respectively. For the Kinsey fantasy ratings,
a one-way ANOVA for age at follow-up was significant, F(2,
22) ⫽ 4.91, p ⫽ .017, while the ANOVA for age at assessment in
childhood approached statistical significance, F(2, 22) ⫽ 2.58, p ⫽
.098. At follow-up, participants classified as bisexual/homosexual
were, on average, significantly older than participants classified as
heterosexual or asexual, t(21) ⫽ ⫺2.54, p ⫽ .019, and t(8) ⫽
⫺2.37, p ⫽ .046, respectively. There was no significant difference
in the mean age at follow-up between participants classified as
heterosexual and those classified as asexual, t(15) ⫽ ⫺1.30, p ⫽
.211. For the Kinsey behavior ratings, the one-way ANOVAs for

SPECIAL SECTION: GENDER IDENTITY DISORDER IN GIRLS

41

Table 4
Means and Standard Deviations of Age (in Years) as a Function of Kinsey Ratings in Fantasy
and Behavior
Exclusively
heterosexual

None
Age

M

SD

M

Bisexual/homosexual

SD

M

SD

p

9.75
27.50

2.73
5.88

.098

10.02
27.45

1.91
6.93

By Kinsey fantasy ratingsa
At assessment
At follow-up

12.42
17.22

.35
.18

7.96
21.76

3.10
4.78

.017

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

By Kinsey behavior ratingsb
At assessment
At follow-up

9.94
20.66

3.99
5.11

7.51
22.81

2.51
4.79

.142
.087

a
For participants grouped by Kinsey fantasy ratings, n ⫽ 2, n ⫽ 15, and n ⫽ 8 for participants with no fantasies,
exclusively heterosexual fantasies, and bisexual/homosexual fantasies, respectively. b For participants grouped
by Kinsey behavior ratings, n ⫽ 8, n ⫽ 11, and n ⫽ 6 for participants with no behaviors, exclusively heterosexual
behaviors, and bisexual/homosexual behaviors, respectively.

age at assessment and follow-up were nonsignificant, F(2, 22) ⫽
2.14, p ⫽ .142, and F(2, 22) ⫽ 2.73, p ⫽ .087, respectively.

z-composite, the main effect for Kinsey ratings in behavior remained statistically significant, F(2, 18) ⫽ 3.58, p ⬍ .05, partial
␩2 ⫽ .29.

Social Desirability
One-way ANCOVAs (age at follow-up covaried) were conducted to evaluate the proportion of socially desirable responses on
the M–C SDS for participants classified as heterosexual, bisexual/
homosexual, and asexual in fantasy and behavior. There were no
significant differences in the proportion of socially desirable responses on the M–C SDS as a function of Kinsey ratings in either
fantasy or behavior, F(2, 20) ⫽ 1.00, ns, and F(2, 20) ⬍ 1,
respectively (data not shown; see footnote 2).

Relation Between Sex-Typed Child Behavior and Sexual
Orientation
To evaluate whether degree of cross-sex-typed behavior in
childhood was related to sexual orientation at follow-up, we used
the z-composite of sex-typed behavior as a function of Kinsey
classification in fantasy (heterosexual, bisexual/homosexual, asexual). With age at follow-up covaried, there was no significant
difference in participants’ cross-sex-typed behavior in childhood
as a function of sexual orientation in fantasy, F(2, 21) ⫽ 1.06,
partial ␩2 ⫽ .09 (data not shown; see footnote 2). For Kinsey
ratings in behavior, however, a one-way ANCOVA was significant, F(2, 21) ⫽ 6.45, p ⫽ .006, the strength of which was large,
as assessed by partial ␩2, with the Kinsey ratings accounting for
37% of the variance of participants’ cross-sex-typed behavior in
childhood. Participants classified as bisexual/homosexual (M ⫽
.52, SD ⫽ .49) had significantly more cross-sex-typed behavior in
childhood than participants classified as heterosexual (M ⫽ ⫺.04,
SD ⫽ .45) or asexual (M ⫽ ⫺.33, SD ⫽ .39), both ps ⬍ .05. There
was no significant difference in the mean z-composite of sex-typed
child behavior between participants classified as heterosexual and
those classified as asexual (see footnote 2).
For the Kinsey ratings in behavior, we reran this analysis with
the 3 gender-dysphoric participants removed (2 were classified as
bisexual/homosexual and 1 was classified as asexual). For the

Relation Between Recalled Childhood Cross-Gender
Behavior and Gender Identity at Follow-Up
We conducted an evaluation of recalled cross-gender behavior
between gender-dysphoric and non-gender-dysphoric participants.
Table 5 shows the means and standard deviations of the RCGI
Factor 1 score. Participants classified as gender dysphoric at
follow-up (n ⫽ 2; Ms ⫽ 1.29 and 1.81, respectively) recalled
significantly more cross-gender identity and role behavior in childhood than participants classified as having no gender dysphoria
Table 5
Mean Factor Scores and Standard Deviations on the Recalled
Childhood Gender Identity/Gender Role Questionnaire (Zucker
et al., 2006) for Gender Identity Status and Sexual Orientation
at Follow-Up
Group
Gender identity status
Gender dysphoric
No gender dysphoria
(Adolescent girls with GID)
Sexual orientationa
Heterosexual
(Heterosexual comparison sample)
Bisexual/homosexual
(Homosexual comparison sample)

M

SD

d

n

1.55
2.69
(2.15)

.36
.59
(.58)

⫺1.96

2
18
(25)

2.82
(3.34)
1.84
(2.68)

.54
(.53)
.44
(.72)

1.88

15
(30)
5
(21)

Note. The absolute range was 1.00 –5.00. A lower score indicates more
recalled atypical gender identity and gender role behavior. Twenty participants completed the questionnaire because the RCGI was not yet part of
the follow-up protocol for 5 participants. Groups and values in parentheses
are from Zucker et al. (2006); the factor scores were from a sample of
heterosexual and homosexual female university students unselected for
gender identity. GID ⫽ gender identity disorder.
a
Sexual orientation was determined on the basis of Kinsey ratings for
fantasy and behavior.

42

DRUMMOND, BRADLEY, PETERSON-BADALI, AND ZUCKER

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(n ⫽ 18; M ⫽ 2.69; range, 1.56 –3.87), t(18) ⫽ ⫺2.62, p ⫽ .017.
As shown in Table 5, the mean Factor 1 score on the RCGI for the
participants with persistent gender dysphoria was more extreme
than it was for a sample of clinic-referred adolescent girls (n ⫽ 25)
with GID reported on by Zucker et al. (2006), whereas the mean
score of the participants without gender dysphoria was somewhat
less extreme.
Further analyses on the RCGI Factor 1 score by sexual orientation revealed that participants classified as bisexual/homosexual
recalled significantly more cross-gender identity and gender role
behavior in childhood than did participants classified as heterosexual or asexual, t(18) ⫽ 3.65, p ⫽ .002.

Discussion
The data reported in this article represent the first systematic
psychosexual follow-up into late adolescence and young adulthood
of clinic-referred girls with potential problems in their gender
identity development. The two key findings were as follows: (a)
the percentage of girls with persistent gender dysphoria was modest but arguably higher than the base rate of GID in the general
population of biological females, and (b) the percentage of girls
who differentiated a later bisexual/homosexual sexual orientation
was moderate but clearly higher than the base rates of bisexual/
homosexual sexual orientation in general survey and epidemiological studies of adolescent girls and young adult women in which
sexual orientation (in fantasy and/or behavior) was assessed with
at least some gradation in response options (as opposed to simple
dichotomous items).
Before providing an analysis of these findings, we note two
limitations of the study. First, the sample size was small, but this
is, at least in part, understandable because the number of referred
girls to specialized gender identity clinics is notably lower than
that of referred boys (e.g., Cohen-Kettenis et al., 2003, 2006).
Second, the present study did not have a concurrent control group
(e.g., a group of girls referred for other kinds of clinical concerns
or a group of nonreferred girls). Accordingly, some of our comparative analyses relied on epidemiological or survey data.
Regarding the persistence of gender dysphoria from the childhood assessment to the follow-up, the present study found that the
vast majority of the girls showed desistance: 88% of the girls did
not report distress about their gender identity at follow-up. The
high rate of desistance appears to differ quite markedly from the
findings of other follow-up studies of adolescent girls and adult
women with GID (in which the baseline assessment is in adolescence or adulthood). In these studies, the rate of GID persistence
appears to be, at minimum, around 70% (Cohen-Kettenis & van
Goozen, 1997; Smith, van Goozen, Kuiper, & Cohen-Kettenis,
2005). In a comparative developmental perspective, then, there
appears to be important variation in GID persistence between
childhood and adolescence/young adulthood.
How might this disjunction be understood? One possibility
pertains to the differences in the DSM criteria for GID that are used
for children versus those that are used for adolescents/adults. The
criteria for GID in girlhood place relatively greater weight on
surface behaviors of cross-gender identification, whereas the criteria in adolescence and adulthood rely more strongly on behaviors
and feelings pertaining to the disjunction between gender subjectivity and somatic sex. Thus, it is conceivable that the childhood

criteria for GID may “scoop in” girls who are at relatively low risk
for adolescent/adult gender dysphoria, which revolves so much
around somatic indicators (e.g., distress regarding breast development or other markers of physical femaleness, etc.).
It should, however, be noted that adolescent girls and adult
women with GID typically recall the same kinds of cross-gender
behavior patterns in girlhood that correspond to the DSM criteria
for GID in childhood (e.g., Blanchard & Freund, 1983; Pearlman,
2006; Zucker et al., 2006), which are then augmented and exacerbated by the external physical markers of biological femaleness
at puberty. Indeed, in the present study, the recalled sex-typed
behavior from childhood of our participants was reasonably similar to the childhood recollections of girls with GID assessed for
the first time in adolescence (see Table 5).
In the present study, 40% of the girls were not judged to have
met the complete DSM criteria for GID at the time of childhood
assessment (although some of these girls likely had met the complete criteria at some earlier point in their development). Thus, on
the one hand, it could be argued that if some of the girls were
subthreshold for GID in childhood, then one might assume that
they would not be at risk for GID in adolescence or adulthood. On
the other hand, it could be argued that cross-gender identification
in girlhood (including subthreshold GID) is a risk factor for later
GID; that is, under some conditions, there is an intensification of
cross-gender identification that results in the development of gender dysphoria (see Green, 2003). Indeed, clinical experience with
adolescent girls with GID indicates that not all of them would have
met the complete criteria for GID in girlhood. Indeed, it is not
uncommon for the parents of these girls to recall that their daughters identified as “tomboys” during childhood and that they did not
remember them voicing the desire to want to become a boy, but
that their gender dysphoria emerged only around the time of
puberty (see, e.g., Pearlman, 2006; Zucker, 2006, Case 1).
If one accepts the argument that girlhood cross-gender identification is a risk factor for gender dysphoria in adolescence and
adulthood, the relatively high rate of desistance in the current study
(in comparison with the relatively high rate of persistence seen in
gender-dysphoric girls and women assessed for the first time in
adolescence or adulthood) suggests that there is some type of
plasticity in gender identity differentiation that operates early in
development but then narrows considerably by adolescence. Thus,
at least among the girls in the present sample, some factor or set of
factors may have operated to lessen the likelihood that their gender
dysphoria or cross-gender identification would persist or intensify
in adolescence and adulthood. Of course, such factors could include both biological and psychosocial influences, but the systematic identification of such factors was beyond the scope of the
present investigation.
To our knowledge, the results of the present study represent the
first prospective data set that shows that girlhood cross-gender
identification is associated with a relatively high rate of bisexual/
homosexual sexual orientation in adolescence and adulthood. Using survey data on sexual orientation in young women as a comparative metric, we estimated that the odds of reporting a bisexual/
homosexual sexual orientation in fantasy was 8.9 –23.1 times
higher in the present sample and that the odds of reporting a
bisexual/homosexual sexual orientation in behavior was 6.7–15.5
times higher. In this respect, the data show at least some conver-


Related documents


PDF Document genderdysphoria
PDF Document 10 1023 a 1010243318426
PDF Document lgbt religious suicide
PDF Document truthknowsnogender
PDF Document dealing with dual diagnosis in stigmatized minorities
PDF Document copy of truth knows no gender final


Related keywords